Journal of Ecology (1988), 76,451 -465
LITTERFALL A N D LEAF DECAY IN THREE
AUSTRALIAN RAINFOREST FORMATIONS
MARGARET D. LOWMAN
Department of Zoology, University of New England, Armidale, Australia
SUMMARY
(1) Litterfall was measured monthly over five years (1979-84) in three representative
rainforest formations of New South Wales. Australia. Known weights of leaf material of
selected canopy tree species were laid out on the forest floor in mesh bays and weighed
monthly to determine relative rates of leaf decay.
(2) Mean annual litterfall was 6.2 (f0.22). 7.3 (k0.57). and 10.0 (*0.74) t ha-' for
cool temperate (microphyll fern forest o r MFF), warm temperate (simple notophyll vine
forest or SNVF), and subtropical (complex notophyll vine forest or CNVF) rainforests,
respectively, with an average of 7.4 1 ha-' for rainforests in New South Wales.
(3) Litter caught in traps was sorted into major components of learnlaterial, wood and
reproductive parts for two years to determine proportions of cach component and to
quantify the seasonal patterns of the canopy. Leaf material averaged 54°A1 over all years
and sites, with 35% wood and 1 lo/;, reproductive parts. M F F exhibited bimodal peaks of
leaf-fall in autumn (March-June) and spring (September-October). while C N V F canopies
showed a summer leaf-fall peak (November-December), and SNVF had an early summer
(Octobel-Decembe~l leaf-tall peak.
(4) Multiple analyses of varlance were performed on the data to analyse a range of
spatial and tcmporal factors affecting littcrfall mcasurcmcnts. In somc cascs, estimates of
litterfall were significantly affected by presence of overhanging subcanopy branchcs,
closeness to tree trunks. and species of canopy Lree overhead. Litterfall did not vary with
placement of traps between rainforest patches, under different sides of one trce, at diffcrcnt
distances away from the trunk of one tree (albeit under its canopy) o r among individuals of
the same canopy species.
(5) Australia11 rainforest trees exhibited valiability in rates of leaf decay. ranging from
less than six months for complete leaf decay in Dendrocnide excelsa to over three years lor
Norltofagus moorei.
(6) The differences In litterfall and leaf decay rates are discussed in relation to
differcnccs in biomass, seasonality, and activities of canopy-associated fauna of these
rainforest formations.
TNTRODUCTTON
Litterfall collection is a standard non-destructive technique for assessing the productivity,
phenology, and turnover of biomass in a forest (Newbould 1967). Litterfall measurements range from 0.6 t h a y 1y y ' in arctic tundra (Levina 1960) to 25.4 t h a y ' y y l in
tropical lowland forests (reviewed in Bray & Gorham 1964; Golley 1978). In particular,
the amount of leaf material falling reflects a forest's productivity. Rainforests, with their
dense leafy canopies, are considered highly productive plant communities, and have
relatively high litterfall (e.g. Proctor et al. 1983).
Australian rainforests are distributed discontinuously down the eastern side of the
mainland from Cape York, Queensland to Cape Otway, Victoria, in areas with at least
1300 mm annual rainfall that extend no further inland than 150 km (Webb 1959; Francis
1981). There are four major rainforest formations in Australia, varying with latitude,
452
Litterfall in Australian rain.forests
altitude, and environmental conditions. Tropical rainforest is restricted to North
Queensland. Subtropical rainforest extends from southern Queensland down to central
New South Wales in warm, moist pockets often associated with rich, basaltic soils.
Temperate rainforests occur throughout New South Wales, Victoria, and Tasmania:
warm temperate forests on sites with lower rainfall and poorer soils than required for
subtropical forests; and cool temperate forests in cooler, often montane, regions. Within
these four general rainforest formations, there are further types defined by botanic or
environmental characteristics (Webb 1959).
The litterfall of Australian tropical rainforests in Queensland was estimated as 9.25 t
ha h ' (Brasell, Unwin & Stocker 1980), and from 7.28 to 10.53 t ha-' y- (Spain 1984).
Similar studies have never been conducted in the other three rainforest formations further
south in Australia, although leaf litter alone has been measured and its mineral content
analysed (Webb et al. 1969). As part of extensive research on leaf growth dynamics of
Australian rainforest canopies (Lowman 1982), litterfall and leaf decay rates of
subtropical, and warm and cool temperate rainforests were measured. Leaf growth and
herbivory levels are reported elsewhere (Lowman 1982, 1984, 1985); leaf senescence and
decomposition, the final stages in the turnover of photosynthetic biomass in forests, are
reported here.
Studies of litterfall underestimate the biomass of a stand, however, if the leaf portions
eaten by herbivores are not also taken into account (Jordan 1971; Lowman 1984). This
failure to account for leafmaterial consumed by herbivores represents only one of several
potential errors in using litterfall studies to estimate productivity. Different published
studies exhibit considerable methodological variation in sizes and materials, and numbers
and placement of traps in sites sampled; duration of collecting period; and categories of
litter sorted (reviewed by Proctor 1983). This makes comparison of data from different
studies difficult. Similarly, measurements of litter decay rates have their associated
methodological problems (Anderson & Swift 1983).
The aim of this study was to quantify and compare litterfall and leaf decay in thrcc of
the four major formations of Australian rainforest. T o place the comparisons on a sound
basis, varialion in resulls from different melhods of using lilter Lraps was crilically
examined, including variation in measurements of litterfall with trap placement and with
month and year.
MATERIALS AND METHODS
Location of rainforests
Litter was collected over a five-year period (March 1979-February 1984) in three
formations of rainforest in New South Wales, Australia, and leaf decay rates were
measured over two years (1980-1982). Sites included New England National Park at 30"
30's (cool temperate or microphyll fern forest, MFF, dominated by Nothofagus moorei F.
Muell.); Royal National Park at 34" 10's and Dorrigo National Park, Never Never
Region at 30" 20's (warm temperate or simple notophyll vine forest, SNVF, including
Ceratopetalum apetalum D. Don, Doryphora sassafras Endl., Ackmena smithii (Puir.)
Merrill et Perry and others in the canopy); and Dorrigo National Park at 30° 20' and Mt
Keira Reserve at 34" 30's (subtropical or complex notophyll vine forest, CNVF, with
Dendrocnide excelsa (Wedd.) Chew, Doryphora sassafras, Sloanea sp., Ficus sp., Orites
excelsa R.Br., and many others in the canopy). Forest classifications follow the
nomenclature of Webb (1 959), who also described the floristic and physical environment
associated with each. Site descriptions are summarized in Table 1, but listed in greater
detail elsewhere (Lowman 1982).
Litterfall measurements
At least twelve litter traps were placed under mature canopies in representative sites of
each of the three types of rainforest formation. Trap placement was random, except in
cases where specific factors of variability in litter trap placement were being tested in
addition to the general study of litterfall. For example, four traps in the C N V F were
placed under different sides of Dendrocnide excelsa to compare differences in litterfall
weights with direction (N, S, E, W of the trunk) of trap placement. In addition, traps were
replicated under the canopies of five species (Nothofagus moorei, Toona australis F . Muell,
Dorjphora sassajras, Ceratopetalum apetalum, and D. excelsa) to quantify their
phenological events. Traps of 1 m2 dimension were constructed of strong nylon mesh
suspended on frames of plastic conduit tubing approximately 0.5 m above ground. Litter
was collected monthly, and litter weights were corrected to thirty-day intervals in cases
where wet weather prevented collection on the thirtieth day. All litter from each trap was
collected in paper bags, and returned immediately to the laboratory, where it was dried
(85 "C for at least 48 h), sorted, redried, and weighed. During the first two years (March
1979-February 1981), litter from each trap was sorted monthly into the components of
wood, reproductive parts (fruits and flowers), and leaves. Wood comprised all branch
sizes, including one event of a large treefall: the section falling across the trap was sawed,
dried and weighed. Small portions of 'trash' that accumulated on the bottoms of traps
were allocated to the 'wood' component, because it appeared to be shattered sections of
woody material; this trash was alwaysc5% of total litter weight, although higher
proportions have been obtained elsewhcrc (Proctor et al. 1983). Leaf material was further
sorted by species to determine the leafing phenologies of certain canopy trees growing in
proximity to each trap. During the third and fourth years (March 1981-February 1983),
litter from each trap was not sorted but was simply dried and weighed each month.
During the fifth year (March 1983-February 1984), litter was collected sporadically, so
TABLE1. Summary of descriptive characteristics of the rainforest sites studied.
Site
New England National Park
Royal National Park
Dorrigo National Park
Never Never Region
Glade Region
Mt Keira Reserve
Forest
type Latitude
Altitude
(m)
Mean annual
rainfall
(mm)
Maximum
canopy
height
Stand
Stand
(m)
density* diversity
MFF
SNVF
30"30'S
34"10'S
1200
20
2000t
1302t
40
33
70
69
12
14
SNVF
CNVF
CNVF
3020's
30'20's
34"30'S
800
800
400
20045
20045
1302f
35
41
42
-
-
65
63
18
15
Number of individual trees ( > 1 m height) and species as measured in a 525 m2 plot, 70 m x 7.5 m (see
Lowman 1986 for illustrations of these profile diagrams).
t Roger Hacking, Ranger, personal communication.
f Records from town of Waterfall meteorological station.
5 Records from town of Dorrigo meteorological station; see Lowman 1982 for monthly weather data at
each site.
454
Litterfall in Australian rain.forests
only the total annual litter per trap was calculated. From these sorting regimes, both
seasonal and annual comparisons of the temporal variation in litterfall were examined.
Comparisorzs of litterfall with diferent trap placements
In addition to seasonal analyses of litterfall, various types of spatial variation in
litterfall were examined:
(a) Large-scale site differences in litterfall, both between different rainforest formations (MFF, SNVF, and CNVF) and between similar rainforest formations (two sites of
SNVF) in different geographic regions of New South Wales.
(b) Small-scale site differences in litterfall (between three pockets of M F F that were
isolated but within 5 km of one another).
(c) Differences in litterfall in relation to positions of nearby trees: (i) placement of litter
traps in relation to distance from tree trunks (1 m, 3 m and 5 m intervals from bole of D .
sassafras in SNVF, Royal); (ii) placement of litter traps in relation to aspect around tree
trunks (N, S, E, W around D . excelsa in CNVF, Dorrigo); and (iii) placement of traps
adjacent to N . moorei trees vs. not adjacent to trees in the cool temperate rain forest
(MFF).
(d) Trap placement in relation to the canopy directly above: (i) closeness of
overhanging branches (low vs. high overhanging branches in the M F F and the CNVF);
(ii) species of tree overhanging a litter trap ( D . excelsa vs. T , alrstralis vs. D . sassafras in
CNVF; D . sassafras vs. N. moorei in MFF); and (iii) traps under different individuals of
one species (three T . australis, three D . sassafras and three D . excelsa trees in the CNVF;
three N . moorei and three D . sassafras trees in the MFF).
Monthly dry weights (g) of litter from each trap were analysed for variability using
analyses of variance and the significant differences ranked by Student-Newman-Keuls
ranking procedure. Cochran's Lest was used prior to analyses to determine homogeneity
of variance; where necessary, data were transformed to log ( x + 1).
Leaf decav
measurement.^
Leaf decay rates were measured for five species: Nothofagus moorei and Doryphora
sassgfras, the dominant canopy and a common understorey species, respectively, in the
f r a sCeratopetalum apetalum, common canopy trees in the SNVF; and
MFF; D . . ~ a . ~ . ~ a and
D . sassafras, Derzdroozide excelsa, and Toona australis in the CNVF canopy. More
detailed information on the leaf growth, herbivory, and abscission in these species is listed
elsewhere (Lowman 1982, 1984).
For each species in each rainforest formation, thirty plastic mesh bags (5 mm2 mesh)
were placed on the forest floor, each containing 30 g fresh weight of old leaves. Leaves
were picked from the basal section of branches in the canopy (i.e. shade leaves), and were
judged to be old by their position on the branch, the presence of epiphylly or of reddening
yellowing pigmentation, and corresponded to 'age class 5' when analysed for leaf
chemicals and toughness (Lowman & Box 1983). Three bags per species per site were
collected each month for eight months; and the last two collections were made at twomonthly intervals so the final (tenth) collection occurred exactly one year after
commencement of the experiment. The leaf material in each mesh bag collected was
transferred to a paper bag, and returned to the laboratory immediately, where it was dried
at 85 "C for 24 h, and weighed. Three bags of each leaf treatment were dried and weighed
at the commencement of the experiment, to serve as the initial sample standard (i.e. 0%
leaf decay). The dry weights were then transformed into percentage loss based upon the
initial sample weights, which represented 100% leaf material.
The entire field design was repeated over two time intervals (commencing in September
1980 and in January 1981) to test for seasonality in the activity of decomposers as
reflected by time of initiation of leaf decay. In addition, both old ( > 2 years old: age class
5) and mature (1 year old; age class 3; Lowman & Box 1983) leaves of C. upetalum were
tested separately and compared in the SNVF (Royal). Species were tested in at least two
sites where possible, both of which represented their natural distribution: T. australis and
D. excelsa in CNVF (Royal) and SNVF (Dorrigo); D. sassafras in M M F (New England),
SNVF (Royal), and CNVF (Dorrigo); C. apetalum in SNVF (Royal); and N. moorei in
M F F (New England). Since shade leaves were used in all cases, it is assumed that the
resulting decay rates are faster than would be obtained for sun leaves, which are tougher
and thicker (Lowman & Box 1983).
RESULTS
Litterfall measurements
Average annual litterfall
The mean total litterfall for all rain forest sites in New South Wales, Australia was 7.4 t
ha-' y-I (Table 2). Total litterfall weights (t ha-l y I) ranged from as high as 10.0
(k0.74) in the CNVF, to 7.3 (+0.57), 5.4 (k0.43) and 6.2 (k0.22) in SNVF (Royal),
SNVF (Dorrigo), and MFF, respectively. The greatest litterfall occurred in the most
complex rainforest type studied (CNVF), which has a higher and more extensive canopy
area than either the SNVF or the M F F (Table 1, see also Lowman 1986). Despite the
relatively long duration of litter collections, total litterfall was nonetheless significantly
different among the four years for each rainforest formation (CNVF: F,1.36 = 7.22; SNVF:
F R ,-.~5.38; MFF: FI 1,36 = 2.73, P < 0.05), probably because different annual rainfall and
TABLE
2. Average dry weights (t h a 1y l ) of litterfall at four rainforest sites in New
South Wales, Australia.
Site
New England
National Park, MMF
Royal National
Park SNVF
Dorrigo National
Park, Never Never
Region, SNVF
Dorrigo National
Park, CNVF
All sites, average
Litterfall component (+S.E.)
Leap
Wood*
Fruit*
(40Litter component)
4.0
(54%)
2.6
(35%)
0.8
(I I%)
Total?
7.4
(100%)
* Calculated from first two years; collection at monthly intervals; litter
components were not sorted during third and fourth years of measurement (n = 24
for each rainforest formation).
t Calculated from four years' collection at monthly intervals sodoes not exactly
represent total of first three columns (n=48 for each rainforest formation).
Litterfall itz Australiatz raitz ,fi~rests
456
storm patterns led to different intensities of wood fall (see Lowman 1982 for more detailed
weather data).
The litterfall components comprised an average of 54% leaf material, 35% wood and
I I% fruit (Table 2). These proportions were consistent for leaf material among the
different sites, but varied slightly for wood and fruit. Woodfall ranged from only 30-33%
in the SNVF sites, to 43% in the M F F whose montane location is characterized by high
winds and storms, resulting in relatively frequent branch and treefall. Fruitfall, however,
was low in the M F F (only 4%), where the predominant canopy tree (Nothofagus moorei)
is a mast-seeder producing reproductive parts only every 3-5 years. This occurred during
spring 1980, whereas fruitfall was negligible in the M F F during the other three years. In
contrast, the CNVF and SNVF litter comprised 13-17(% fruit material; these sites have
many canopy species that reproduce annually, some with heavy, fleshy fruits.
Mean leaf-fall was 4 t h a ' y-' for all sites, but ranged from as high as 5.4 tin the CNVF
to less than half that (2.09 t) in the SNVF site only 10 km away on the edge of the
escarpment. Mean woodfall was 2.6 t ha-' y-' although its associated standard errors
were relatively high. Trunk or large branchfall into or across the traps, albeit relatively
infrequent, did occur and resulted in high variability between monthly collections.
Woodfall was highest in the CNVF, at 3.2 t ha-' y-!, and lowest in the Dorrigo SNVF
site, with only 1.8 t. The proportion of wood was highest in the MFF, however, where the
annual fall of 3.0 t comprised 43% of total litterfall. discussed above. Fruitfall averaged
0.8 t ha-' y-' in all New South Wales sites sampled, ranging from 1.4 t in the CNVF to
only 0.3 t in the MFF.
Litterfall weights were compared within and between the three main types of rainforest
formation: MFF, SNVF, and CNVF in increasing order of complexity (Table 3).
Litterfall and each of its three components-leaf,
wood and fruit-were
highly
significantly different ( P <0.001) between the three formations, and were highest in the
CNVF in all four litter components. Within the sites, however, total litter and wood were
not significantly different between traps at three of the four rainforests analysed. In
contrast, fruit and leaf litter varied significantly between traps in three out of four sites,
most likely a consequence of trap placement: fruit and leaf-fall varied according to tree
species above each trap, whereas wood and total litter were fairly constant between traps.
TABLE
3. Comparisons of litterfall (total litter, leaf, wood, and fruit) (t h a 1 )within
and between four rainforest regions in New South Wales, Australia. See text for
explanation of site initials.
MFF
Litter component
Total litter
Leaf
Wood
Fruit
1.21 N.S.
22,96***
0.94 N.S.
6.37..
N.S. non-significant.
* P < 0.05.
** P<0.01.
*** P<0.001.
Within sites
SNVF
SNVF
(Royal)
(Dorrigo)
15,30***
63.80***
408.80***
11.07**
3.86
0.1 1
1.17
0.49
N.S.
N.S.
N.S.
N.S.
CNVF
Between
sites
1.19 N.S.
12,46***
0.83 N.S.
18.11***
40.44***
23.23***
19.37***
28,61***
CNVF
SNVF (Royol)
T
l
M
A
M
J
J
A
I
Z
t
U
1
N
I
D
I
J
4
F
0 ~ " " ' " " " '
M
A
M
J
J
A
S
O
N
D
J
F
Time (calendar month)
FIG. I . Litterfall (mean weight per site* 1 S.E. g m 2 y ' ) in four rainforest sites, New South
Wales. Australia: MFF, m~crophyllfern forest; SNVF (Royal) and SNVF (Dorrigo), simple
notophyll vine forests in Royal and Dorrigo National Parks; and CNVF, complex notophyll vine
forest. Each po~ntrepresents a monthly mean calculated from four (MFF, CNVF) or three
(SNVF) years' litter collections.
Seasonal fluctuations in litterfall
All four rainforests had bimodal peaks in their annual patterns of litterfall, but with
different seasonal fluctuations (Fig. la-d). In the MFF, litterfall peaked during autumn
(May) and sprlng (September-October), while in both SNVF sites, the maximum fall
occurred during spring (October) with a lesser summer peak in November-December
(Royal) or February (Dorrigo). The CNVF exhibited highest litterfall during early
summer (December), with a smaller peak during autumn (May).
The seasonality ofeach component of litterfall was also quantified, but only over a twoyear span (Fig. 2 a d ) . These graphs reveal that not all of the total litterfall peaks (Fig. 1 ad) were composed of consistent proportions of leaf, wood and reproductive matter.
Furthermore, the peaks composed of wood (as a result of a tree or large branchfall) were
not annually repeated patterns, but random events.
In the MFF, leaf-fall was highest during autumn (May-June) and during spring
(September-October) (Fig. 2a). Because the M F F canopy is predominantly Nothofagus
moorei, the litter patterns closely parallel the phenology of this species. This bimodal
pattern of leaf-fall is typical of many evergreen temperate trees: a portion of the canopy
abscises immediately before the winter months, and another portion falls simultaneously
as new leaves flush in the spring. In total, approximately half the canopy was shed
annually, leaf longevity of N. moorei being approximately two years (Lowman 1982). The
high woodfall in May reflects the heavy rains and storms that consistently occurred
458
Litterfall in Australian rain forests
during autumn, causing both tree and large branchfalls into the litter traps. Fruitfall was
relatively low in the MFF. The majority of reproductive parts collected in the litter traps
were from vines, epiphytes and understorey such as Elaeocarpus holopetalus F. Muell. and
D. sassafras. The peak during September-October, however, resulted from the flowering
and fruiting of beech during 1980. As a mast-seeder, beech trees only flower every three to
five years; the reproductive litter weights of beech were negligible in other years.
Both SNVF sites had similar patterns of spring leaf-fall (Fig. 2 b,c) but the peaks were
slightly later in the more southerly site (October-December in Royal as compared to
September-October in Dorrigo, 500 km to the north). Leaf abscission times coincided
with the months of major leaf-flushing activities of most canopy trees in these rainforest
sites (Lowman 1982). The shedding of bark and branches was heaviest during spring and
summer, with very little wood falling during the autumn and winter. N o large treefalls
occurred in the vicinity of the SNVF litter traps during 1979-81, so the annual pattern was
fairly homogeneous. Fruitfall peaked in August at Royal (due in part to the successful
flowering in 1980 of lilly pilly, Acmena smithii, which has a heavy, fleshy fruit). Fruitfall in
Dorrigo peaked during February, here again reflecting the phenology of a canopy species
with large, fleshy fruits (Endiandra introrsa C.T. White, Dorrigo plum).
The seasonal patterns of leaf-fall in the CNVF (Fig. 2d) peaked during spring-summer
(September-December), a time which coincided with the leaf flush of most canopy trees.
The smaller leaf peak in May was due to a deciduous species Toona australis that lost its
entire canopy during autumn. (More information on leaf-fall of individual species whose
canopies were above the litter traps is given below.) Woodfall peaked in May, a result of
the heavy rainstorms that caused Lree and large branchfalls (the same storms that affected
woodfall in MFF). Woodfall throughout the rest of the year consisted mainly of small
branches and bark flecks. Litterfall of reproductive parts was sporadic (albeit higher than
in any other rainforest formation), with peaks during periods of fruitfall of species with
fleshy fruits. They included Endiandra introrsa during February-March; Brachyclliton
aceri/olium (A. Cunn. ex Don) F. Muell. (flowers) during November-January; 7oona
australis in March; Cryptocarya glaucescens R.Br., Geoissos benthamii F. Muell. and
Sloanea australis (Benth.) F. Muell. in April; C . apetalum (flowers) and Ficus macrophylla
Desf. ex Planch. fruits in January; D. sassa/ras (flowers) in June; and Acmena smithii fruits
in August.
Statistical differences in litterfall weights were tested between the four seasons (analyses
of variance among months were not valid, because monthly trap weights were not
independent), using months as replicates of a given season. At all sites, litterfall was
significantly highest in the spring and lowest in the winter (MMF, F3,320=11.09; SNVF,
fi,320= 58.17; and CNVF, F 3 , 3 2 0 = 46.85, P < 0.001).
Comparisons of litterfall in relation to placement of litter traps
Nearness to tree trunks. Litter weights from traps situated on different sides (N, S, E, W)
of trees of D. excelsa in the CNFV were statistically similar for all litter components. In
the closed canopy environment of the CNVF, prevailing winds did not appear to alter the
direction of falling litter. Secondly, traps situated at different distances from tree trunks ( I
m, 3 m, 5 m away from D. sassafras in SNVF) showed no significant difference in weights
of any litter component. The trap nearest to the tree trunk did not accumulate heavier
litter weights from the dense canopy above, nor did litter appear to be deflected away from
the trap by the nearby trunk. Thirdly, litterfall was compared between traps placed
adjacent to tree trunks ( N . moorei) and traps placed in the open away from tree trunks,
Litterfall in Australian rain forests
albeit within the same forest canopy. Both woodfall and N. moorei leaf-fall were
homogeneous in this test, but fruit and total leaf-fall varied significantly with distance
from the trunk. The heaviest (wood) and most homogeneous (N. moorei leaves) litter
components were unaffected by trap pIacement in relation to tree trunks, but fruit and
leaves of other species fell more heavily into traps located away from tree trunks.
Canopy above trap. Litter traps placed under low, overhanging subcanopy branches
showed little significant difference as compared to traps not under low overhanging
branches. In the MFF, all components of litterfall were homogeneous whether the trap
was situated under subcanopy branches or not, (F1.360=2-66,N.S.). In the CNVF,
however, leaves and fruits fell homogeneously, regardless of presence or absence of
6 ; N.S.); but wood was
overhanging subcanopy branches (Fl,168=0'15;F 1 , 1 ~ ~ = 3 . 4both
deflected so that significantly less fell into traps beneath overhanging branches
(F1.16~
= 10.15, P < 0.01) as compared to traps under open subcanopy.
Litterfall varied markedly in traps under different species of canopy trees. In the MFF,
traps under D. sassafras collected less total litter than under N. moorei (F1,264=8.50,
P<0.01). The leaf-fall of each individual species was statistically higher under its own
canopy, but the components of fruit and wood were homogeneous. In the CNVF, wood
weights were homogeneous between traps placed under the canopies of three species; but
fruit, leaves, and total litter varied significantly under different canopy trees.
In most cases, litterfall did not differ significantly among replicate individuals of one
canopy tree species. The only exception of this was D. excelsa, where all litter components
except leaf-fall varied between individuals. These differences probably reflect litterfall of
the other trees growing adjacent to the traps since litter components of D. excelsa are
extremely light, containing proportionally more water weight than for any other
rainforest tree measured (Lowman 1982).
In summary, litter traps are inevitably situated under different canopy conditions.
Litter weights in the M F F were not affected by presence or absence of subcanopy
branches or differences between individual trees, but varied between traps under canopy
(N. moorei) or subcanopy (e.g. D. sassafras) individuals. In the CNVF, litterfall weights
varied slightly in relation to presence of overhanging subcanopy branches; and under
different species and individuals of canopy trees. The increased complexity of the CNVF
appears to require a greater variability in trap placement to assure representative litter
collection.
Leaf decay measurements
Leaf decay rates varied considerably among species ranging from as little as four
months for D. excelsa, to over two years for N. moorei (Fig. 3).
D. excelsa leaves decayed very rapidly (Fig. 3a), with over 95% of the original dry
weight lost after only four months. The remaining 5% consisted mainly of midveins that
required up to eight months to break down. This was the fastest decay rate measured in
this study, but perhaps predictable due to the soft, mesomorphic tissue of D. excelsa
leaves.
Although T. australis leaves were not sclerophyllous as are many Australian rainforest
trees, they required over one year to decay fully (Fig. 3b). Leaves on the forest floor of the
CNVF decayed slightly faster than at the SNVF Royal, probably due to moister, warmer
conditions of the former.
C. apezalum leaves also decayed very slowly, requiring almost one year to lose 50% of
their initial weight (Fig. 3d). The young leaves (1-year old) decayed slightly faster than did
M. D. LOWMAN
D. excelso
D. sossofros
~~~
-
20
10
0 S O N D J F M A M J
J
A
CNVF
l
I
S
1
O
1
N
D
1
J
1
F
1
M
1
A
M
l
J
1
J
~
~
1
A
T i m e (colendor month1
FIG.3. Leaf decay of five rainforest tree species over one year expressed as percentage remaining
of original leaf weight. Sites abbreviated as follows: MMF (mossy microphyll forest); SNVF
(simple notophyll vine forest); and CNVF (complex notophyll vine forest); bars =S.E.
the old leaves ( 2 2 years old); but even they were much slower than mature D. excelsa
leaves.
D. sassafras grew in all three rain forest formations studied, so leafdecay was measured
at each site. Its leaves decayed faster in the CNVF habitat than in the MFF, with an
intermediate rate in the SNVF, Royal (Fig. 3e). Decay was negligible in the M F F floor
during winter, and rates were fastest during summer at all three sites.
462
Litterfall in Australian rain forests
N. rnooreileaves decayed extremely slowly (Fig. 3c), with over 60(%undecayed after one
year. Since beech leaf-fall peaked during two different seasons, it was hypothesized that
the time of leaf-fall may also affect the decay rates (due perhaps to seasonality of
decomposers, temperature, moisture, or to the physical state of leaf tissue at time of
abscission). To test for this, the beech decay experiment was conducted over two time
intervals, one initiated in August and one in January. A two-way analysis of variance
showed differences with respect to both factors. Different months showed significantly
different rates of decay ( F = 8.77, P < 0.001). with summer months exhibiting the fastest
rates; the time of initiation ofdecay was even more significant ( F = 55.64, P < 0.001), with
leaves placed on the forest floor in summer (January) disappearing more rapidly than in
winter (August). The significant interaction ( F = 16.24, P<0.001) implied that a
combination of time of decay initiation and given monthly intervals of decay contributed
a large portion to the variability.
DISCUSSION
Variability in litter production
Rainforest litter usually averages between 6 and 12 t h a 1 y-I (Bray & Gorham 1964;
Golley 1978; Proctor et al. 1983), and this study is no exception. The highest litterfall
(10.0 t ha-' y-I) occurred in the subtropical CNVF, and the lowest in the M F F or cool
temperate rainforest (6.2 t ha-' y-I). Most litter studies in Australian forests have been
conducted under eucalypt or dry sclerophyll canopies where litterfall is approximately 35 t h a - ' y-I (e.g. Ashton 1975; Congdon 1979). In contrast, subtropical Australian
mangroves shed an average of 5.8 t ha-' y-' (Goulter & Alloway 1979). A tropical
rainforest litter study conducted previously in Australia averaged 9.25 t ha-I y-' (Brasell,
Unwin & Stocker 1980). This is surprisingly lower than the litterfall collected in the
subtropical forests of this study, but their tropical sites included some Arauraria
cunninghanlii Ait. ex D. Don plantations. and some undisturbed rainforests that were
actually belts (c. 100 m wide) adjacent to the plantations. Both these factors may have
resulted in litterfall that was slightly lowcr than in a well-developed, undisturbed Lropical
rainforest. In anothcr study, Spain (1984) measured 7.28-10.53 1 h a - ' y-' in three
Australian tropical rainforest sites. The fact that both of these litterfall studies in
Australian tropical rainforests did not have proportionally more litter than in the
subtropical and temperate rainforests reported here may be due to the inclusion of large
branch and treefall weighted in this study; they are often omitted in litterfall studies. The
only extensive litterfall collection in cool temperate rainforests reported weights of 5000
lbs acre-' (5 t ha-') in New Zealand (Miller & Hurst 1957), similar to this study.
When comparing litterfall among different studies, it is often difficult to account for the
potential errors arising from variation in methodology (Proctor 1983). Litter-trap designs
(Newbould 1967) vary in size (from < 20 cm up to 1 m diameter) (shape from round to
square), position (from ground level to > 1 m above ground), construction (e.g. plastic
ice-cream tubs, wooden boxes, plastic mesh bags suspended on frames), and in function
(water proof, animal proof, exclusive to large branches, or inclusive of small reproductive
parts). Second, the number and placement of traps greatly affects the reliability of
litterfall, as shown in this study. Other reviews oflitterfall in the literature have mentioned
the need for confidence limits on litterfall results to indicate the variability of averages
obtained (Proctor et al. 1983). Thirdly, the frequency of collection and duration of study
greatly affect the reliability of results obtained. Even in this study, litterfall was
significantly variable between four years, probably due to the large differences in wood
weights between the years (Table 2). Studies of one year or less simply cannot hope to
estimate litter accurately, because of the annual variation in biological (e.g. fruiting and
flowering) and physical (e.g. rainfall) factors.
Relationships between leaf litter and herbiz1ory
The amount of leaf material in the litterfall was approximately 54%).which is similar to
most other studies, although up to 85% leaf material has been reported in montane
rainforests of Jamaica (Tanner 1980). Leaf-fall, when taken in conjunction with leaf
longevity, can be used to estimate the canopy biomass, but only if the amounts of leaf
surface missing to herbivores have been quantified. Previous studies on defoliation in
these rainforest sites estimated leaf area losses at 14.6% in CNVF, 22.0% in SNVF and
26% in M F F (Lowman 1984). Tf these portions missing to herbivores are added to the
litterfall weights, then the estimates of forest biomass are higher: 10.9 t ha-' y ' in the
CNVF, and 8.5 t h a ' y-' mean biomass for all rainforests measured in N.S.W., Australia
(Table 4). Wood borers and frugivores may also contribute to underestimates of the
weights of both wood and reproductive component as measured by the litterfall.
Most other litterfall studies fail to account for portions of leaves missing to herbivores
(but see Proctor et 01. 1983). When litterfall studies are conducted solely to measure the
biomass entering the decay pathway on the forest floor, this is acceptable. But if, as with
many studies, litterfall measurements are subsequently extrapolated to estimate forest
productivity or above-ground biomass or nutrient availability, then the results may
underestimate the original canopy present. Further studies on wood and fruit losses to
their associated predators before senescence would further quantify the accuracy of
measurements of the turnover of plant material through the litterfall pathway.
Leuf decay in Azcstralian rainforests
Decay represents the final phase in the life cycle of a leaf, whereby its constituents are
recycled back into the forest system. As with litterfall studies, the methodology for leaf
decay studies is extremely variable although relatively few studies have been conducted
(Anderson & Swift 1983). Some investigators tether individual leaves to stakes on the
forest floor (e.g. Gong & Ong 1983), others lay out disks of leaf material (e.g. Heath &
Arnold 1966), and some use mesh bags (e.g. Shanks & Olsen 1961 ; Ewe1 1976; this study).
TABLE
4. Litterfall (t h a ' y - - I ) at four rainforests in New South Wales, Australia
after correcting for leaf area losses to herbivores.
~
Site*
~
- ,
~
b Leafi litter
~ weights
~
~
Corrected
(% leaf area loss) Uncorrected
MFF
SNVF (Royal)
SNVF (Dorrigo)
CNVF
All sites, 2
("/;I of total)
* See text for site abbreviations.
t See Lowman (1984) for calculations of herbivory levels
~ Total
t litter weights
Uncorrected Corrected
464
Litterfall in Australian rain forests
The rates of decay measured in this study were much lower than those for most
rainforest trees in the literature, where half-lives of approximately seventy days are
reported (Ewel 1976). And even slower decay rates would have been obtained if the leaves
had been selected from the upper (sun) rather than the lower (shade) regions of tree
canopies, because sun leaves decay more slowly than shade leaves (Heath & Arnold 1966).
Australian rainforest trees are mainly sclerophyllous, however, with N. moorei, D.
sassafras and C. apetalum all being very tough (Lowman & Box 1983) compared to the
mesomorphic leaves of D. excelsa. The decay rates appeared to correspond to leaf
toughness and tannins, with N. moorei (both tough and high in tannins) having the
slowest decay and D. excelsa (both soft and low in tannins) decaying in less than six
months (Lowman & Box 1983). The relatively slow recycling of most leaf material
through the decomposer pathway may serve to balance the more rapid re-entry of leaf
nutrients through frass (Nicholson, Bocock & Heal 1966), particularly in the M F F
rainforests where the beech leaves suffer relatively high defoliation (c. 31 %) (Lowman
1982; Selman & Lowman 1983) but exhibit slow decay on the forest floor. One of the few
studies to measure frass decay rates showed a half-life of one year for faecal pellets of
millipedes that ate hazel leaves (Corylus avellana) (Nicholson, Bocock & Heal 1966). If the
frass of beech herbivores is similar, then the leaf portions passing through herbivores are
re-cycling at least twice as fast as leaf material that abscissed. In view of the slow leaf decay
rates measured, further studies comparing the recycling of leaf nutrients via direct
consumption (herbivores) and indirect consumption (decomposers) would be of greater
interest in understanding the importance of litterfall in these rainforests.
ACKNOWLEDGMENTS
I am grateful to Margaret French, AnnaMarie Hatcher, Patti Schmitt, W. Higgins, H.
Caffey, D. Lowman and members of Earthwatch ('The Surviving Rainforest') Center for
Field Research for assistance in collecting litter; to P. Myerscough, A. Underwood, H.
Heatwole and A. Burgess for advice during Lhe fieldwork, analysis and preparation of this
manuscript; to the National Parks of New South Wales for permission to conduct this
research; to Sydney University and the Australian Research Grants scheme for financial
support and to Sandra Higgins for typing this manuscript.
REFERENCES
Anderson, J. M. & Swift, M. J. (1983). Decomposition in tropical forests. Tropical Rain Forest: Ecology and
Management (Ed by S. L. Sutton, T. C. Whitmore & A. C. Chadwlck), pp. 287-31 1. Blackwell Scientific
Publications, Oxford.
Ashton, D. H. (1975). Studies of litterfall in Eucalyptus regnans forests. Australian Journal oJBotuny, 23,413433.
Brasell, H. M., Unwin, E. L. & Stocker, G. C. (1980). The quality, temporal distribution and mineral-element
content of litterfall in two forest types at two sites in tropical Australia. Journal oJEcology. 68, 123-139.
Bray, J . R. & Gorman, E. (1964).Litter production in forests of the world. Adt~ancesin Ecologrcol Research, 2,
101-157.
Congdon, R. A. (1979). Litterfall of the paperback tree (Melaleuca curicularis) in the marshes of the Blackwood
River Estuary, Western Australia. Ausrralian Journal oJEcology, 4, 41 1-417.
Ewel, J. J. (1976). Litter fall and leaf decomposition in a tropical forest succession in eastern Guatemala. Journal
oJEcology, 64,293-308.
Francis, W. D. (1981). Australian Rain-Forest Trees. Australian Government Publishing Service, Canberra.
Golley, F. B. (1978). Forest leaf production. The Ecology ofArhorea1 Folivores (Ed by G . G . Montgomery), pp.
17-22. Smithsonian Institution. Washington.
Gong, W. K. & Ong, J. E. (1983). Litter production and decomposition in a coastal hill dipterocarp forest.
Tropical Rain Forest: Ecology and Management (Ed by S. L. Sutton, T. C . Whitmore & A . C. Chadwick),
pp. 275-287. Blackwell Scientific Publications, Oxford.
Goulter, P. F. E. & AUoway. W. G. (1979). Litter fall and decomposition in a mangrove stand, Aric.ennia n~arina
(Forsk.) Vierh.. in Middle Harbour. Sydney. Ausrruliun Journul of Murine and Freshwater Research, 30,
541-546.
Heath, G . W. & Arnold, M. K. (1966). Studies in leaf-litter breakdown. 11. Breakdown rate of 'sun' and shade'
leaves. Pedobiologia, 6, 238-243.
Jordan, C. F. (1971). Productivity of a tropical rain forest and its relation to a world pattern of energy storage.
Journal of Ecology, 59, 127-142.
Levina, V. I. (1960). Determination of the amount of annual litter fall in two types of pine forest of the Kola
Peninsula. Bot. Zh. 45,418423 (cited in Forestry Abstracts 21, 532).
Lowman, M. D. (1982). Leaf growth dynamics and herbicory in Aus~ralianrain forest canopies. Ph.D. thesis.
University of Sydney.
Lowman, M. D. (1984). An assessment of techniques for measuring herbivory: Is rainforest defoliation more
intense than we thought? Biotropica, 16, 264268.
Lowman, M. D. (1985). Temporal and spatial variability in insect grazing of the canopies of five Australian
rainforest tree species. Australian Journal of Ecology, 10, 7-24.
Lowman, M. D. (1986). Light interception and its relation to structural differences in three Australian rain forest
canopies. Australian Journal of Ecology, 11, 163-170.
Lowman, M. D. & Box, J. R. (1983). Variation in leaf toughness and phenolic content among five species of
Australian rain forest trees. Australian Journal of Ecology, 8, 17-25.
Miller, R. B. & Hurst, F. B. (1957). The quantity and nutrient content ofhard beech litter. New Zealand Forestry
Research Notes, 8.
Newbould. P. J. (1967). Methods for Estiniating the Primary Production of Forests. IBP Handbook No. 2.
Blackwell Scientific Publications, Oxford.
Nicholson, P. B., Bocock, K. L. & Heal. 0 . W. (1966). Studies on the decomposition of the faecal pellets of a
millipede (Glomenrs murginata (Villers)). Journal of Ecology, 54, 755-766.
Proctor, J. (1983). Tropical forest litterfall. 1. Problems of data comparison. Tropical Rain Forest: Ecology and
Managemenr (Ed by S. L. Sutton, T. C. Whitmore & A . C. Chadwick), pp. 267-273. Blackwell Scientific
Publications, Oxford.
Proctor, J., Anderson, J. M., Fogden, S . C. L. & Vallack, H. W. (1983). Ecological studies of four contrasting
lowland rain forests in Gunung Mulu National Park. Sarawak. Journul qfEcology 71. 261L263.
Shanks, R. E. &Olson, J. S. (1961). First-year breakdown of leaf litter in southern Appalachian forests. Science,
134, 194195.
Selman, B. & Lowman, M. D. (1983). The biology and herbivory rates of Nooacastriu nothofugi Selman
(Coleoptera: Chrysomelidae), a new genus and species of Nothofag~rsmoorei in Australian temperate rain
forests. Austruliun Journul of Zoology, 31, 179-91.
Spain, A. V. (1984). Litterfall and the standing crop of litter in three tropical Austral~anrainforests. Journal of
Ecology, 72, 947-6 I.
Tanner, E. V. J. (1980). Litterfall in montane rain forests of Jamaica and its relation to climate. Journal of
Ecology, 68, 833-848.
Webb, L. J. (1959). A physiognomic classification of Australian rain forests. Journul of Ecology. 47, 551-570.
Webb, L. J., Tracey, J. G., Williams, W. T. & Lance, G. N. (1969). The pattern of mineral return in leaf litter of
three subtropical Australian forests. Australian Forestry, 33, 99-1 10.
(Received 10 April 1987)